Re:You Can't Sacrifice Partition Tolerance (response to Ston...
by johnrice on Friday October 22, 2010 @09:15PM
(#1822180)
Stonebraker says real systems can obviously achieve "CA" when one node is partitioned off. But he slyly switched out the theorem's definition of "available" for a more colloquial definition. The CAP theorem definition requires every live node, including the lone segregated guy, to respond to clients. Real "CA" systems don't do that--they're "available" because the non-partitioned nodes are still responding.
[ Parent
| on HN ]
Re:You Can't Sacrifice Partition Tolerance (response to Ston...
by RickHull on Friday October 22, 2010 @10:54PM
(#1822423)
> Real "CA" systems don't do that--they're "available" because the non-partitioned nodes are still responding.
This sounds like a fencepost type of error to me. When you have a partition, you are splitting a single set into two subsets. It may be live/dead or live-net1/live-net2. The point is, in the presence of a partition, there are no non-partitioned nodes.
Re:You Can't Sacrifice Partition Tolerance (response to Ston...
by tbrownaw on Friday October 22, 2010 @09:31PM
(#1822216)
The CAP theorem definition requires every live node, including the lone segregated guy, to respond to clients.
It also models nuked (hardware failure, power outage, whatever) nodes as being in an isolated single-node partition. Which becomes rather interesting in combination with that definition of availability.